working

burstpdf.py

@@ -33,7 +33,7 @@ def pdftopil():
     return pil_images
 
 def save_images(pil_images):
-    d = 1
+    d = 0
     for image in pil_images:
         image.save(("split/input%d"%d) + ".jpg")
         d += 1

mirror_crop.py

@@ -14,7 +14,7 @@ while True:
         print("cropping even")
 
         # left, up, right, bottom
-        border = (0, 0, 68, 0)
+        border = (0, 0, 65, 0)
         finalpage = ImageOps.crop(page, border)
         finalpage.save('cropped/page%i.jpg'%i)
 
@@ -23,7 +23,7 @@ while True:
         print("cropping odd")
 
         # left, up, right, bottom
-        border = (68, 0, 0, 0)
+        border = (65, 0, 0, 0)
         finalpage = ImageOps.crop(page, border)
         finalpage.save('cropped/page%i.jpg'%i)
 

remove.sh

@@ -0,0 +1,5 @@
+cd split
+pwd
+rm page0.jpg
+rm -ltr | tail -1
+rm .DS_Store

tesseract_ocr.py

@@ -7,7 +7,7 @@ i = 1
 
 while True:
     try:
-        img = Image.open("cropped/page%i.jpg"%i)
+        img = Image.open("bounding_box/input%i.jpg"%i)
         print(img)
         pdf = pytesseract.image_to_pdf_or_hocr(img, lang="eng", extension='pdf')
         time.sleep(1)

workshop_stream.sh

@@ -1,12 +1,13 @@
-mkdir split
-mkdir rotated
-mkdir ocred
-mkdir bounding_box
-mkdir cropped
 ./merge_scans.sh
+mkdir split
 python3 burstpdf.py
+mkdir rotated
 python3 rotation.py
+mkdir bounding_box
 python3 bounding_box.py
-python3 mirror_crop.py
+mkdir cropped
+# python3 mirror_crop.py
+mkdir ocred
 python3 tesseract_ocr.py
+./remove.sh
 ./merge_files.sh